Closed josephrocca closed 1 year ago
Hi @josephrocca, decapoda-research/llama-7b-hf
hasn't been updated to use the new class name (LLaMATokenizer
-> LlamaTokenizer
), which was later finalized in HF Transformers v4.28.0. You could just pick another LLaMA model from Hugging Face Hub.
Related: https://github.com/hyperonym/basaran/issues/57#issuecomment-1509036822
Running this command:
Gives:
It worked fine with the
cerebras/Cerebras-GPT-1.3B
andbigscience/bloom-560m
models.(Thanks for your work on this project! Currently integrating it with OpenCharacters)